pfnstudio 0.7.0__tar.gz

pfnstudio-0.7.0/.gitignore

@@ -0,0 +1,68 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# Virtual environments
+.venv/
+.deps-venv/
+venv/
+ENV/
+env/
+
+# Editable-install metadata
+_editable_impl_*.pth
+
+# IDE / editor
+.vscode/
+.idea/
+*.swp
+*.swo
+.DS_Store
+
+# pytest / coverage / mypy
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+.coverage
+.coverage.*
+htmlcov/
+.tox/
+.nox/
+
+# Jupyter
+.ipynb_checkpoints/
+
+# Build artifacts
+checkpoint/
+runs/*/checkpoint/
+runs/*/results.json
+runs/*/metrics.jsonl
+*.pt
+*.pth
+
+# Dataset cache (downloaded benchmark datasets)
+.priorstudio/datasets/
+
+# Logs
+logs/
+*.log

pfnstudio-0.7.0/PKG-INFO

@@ -0,0 +1,99 @@
+Metadata-Version: 2.4
+Name: pfnstudio
+Version: 0.7.0
+Summary: CLI for the PFN Studio framework — scaffold, validate, lint, and run prior-fitted foundation model projects.
+Project-URL: Homepage, https://pfnstudio.com
+Project-URL: Repository, https://github.com/profitopsai/pfnstudio
+Project-URL: Issues, https://github.com/profitopsai/pfnstudio/issues
+Project-URL: Documentation, https://github.com/profitopsai/pfnstudio/tree/main/docs
+Author: PFN Studio contributors
+License: Apache-2.0
+Keywords: causal-inference,foundation-models,ml,pfn,prior-fitted-networks
+Classifier: Development Status :: 3 - Alpha
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.10
+Requires-Dist: jsonschema>=4.21
+Requires-Dist: pfnstudio-core>=0.4.0
+Requires-Dist: pyyaml>=6.0
+Requires-Dist: requests>=2.31
+Requires-Dist: rich>=13.7
+Requires-Dist: typer>=0.12
+Provides-Extra: all
+Requires-Dist: huggingface-hub>=0.20; extra == 'all'
+Requires-Dist: mlflow>=2.10; extra == 'all'
+Requires-Dist: modal>=0.62; extra == 'all'
+Requires-Dist: pfnstudio-core[torch]; extra == 'all'
+Requires-Dist: pfnstudio-studio>=0.4.0; extra == 'all'
+Requires-Dist: runpod>=1.6; extra == 'all'
+Requires-Dist: wandb>=0.16; extra == 'all'
+Provides-Extra: dev
+Requires-Dist: pytest-cov>=5.0; extra == 'dev'
+Requires-Dist: pytest>=8.0; extra == 'dev'
+Requires-Dist: ruff>=0.4; extra == 'dev'
+Provides-Extra: hf
+Requires-Dist: huggingface-hub>=0.20; extra == 'hf'
+Provides-Extra: mlflow
+Requires-Dist: mlflow>=2.10; extra == 'mlflow'
+Provides-Extra: modal
+Requires-Dist: modal>=0.62; extra == 'modal'
+Provides-Extra: runpod
+Requires-Dist: runpod>=1.6; extra == 'runpod'
+Provides-Extra: studio
+Requires-Dist: pfnstudio-studio>=0.4.0; extra == 'studio'
+Provides-Extra: torch
+Requires-Dist: pfnstudio-core[torch]; extra == 'torch'
+Provides-Extra: wandb
+Requires-Dist: wandb>=0.16; extra == 'wandb'
+Description-Content-Type: text/markdown
+
+# priorstudio — CLI
+
+The command-line interface for [PFN Studio](https://github.com/profitopsai/pfnstudio),
+the toolkit for training [prior-fitted foundation models](https://arxiv.org/abs/2112.10510).
+
+## Install
+
+```bash
+pip install priorstudio
+```
+
+For training (requires PyTorch):
+
+```bash
+pip install "priorstudio[torch]"
+```
+
+## Commands
+
+```text
+priorstudio init <dir>            # scaffold a new FM project
+priorstudio validate <path>       # check artifacts against JSON Schema
+priorstudio lint <project>        # cross-reference + style checks
+priorstudio sample <prior.yaml>   # draw N tasks from a prior
+priorstudio run <run.yaml>        # execute a training run end-to-end
+priorstudio predict <run-dir>     # inference against a trained checkpoint
+priorstudio export <project>     # tar-gzipped project archive
+```
+
+Run `priorstudio --help` for the full list and `<cmd> --help` for each
+subcommand's flags.
+
+## What this CLI is for
+
+PFN Studio organises every PFN project around five first-class
+artifacts: **priors** (synthetic data generators), **models** (block
+compositions), **evals** (benchmarks + metrics), **runs** (training
+manifests), and **initiatives** (research workstreams). This CLI
+operates on the file layout those artifacts produce — scaffolding new
+projects, validating them, running training, and exporting them for
+sharing.
+
+The full story (concepts, architecture, examples, marketplace catalog)
+lives at the main repo:
+**[github.com/profitopsai/pfnstudio](https://github.com/profitopsai/pfnstudio)**
+
+## License
+
+Apache-2.0.

pfnstudio-0.7.0/README.md

@@ -0,0 +1,49 @@
+# priorstudio — CLI
+
+The command-line interface for [PFN Studio](https://github.com/profitopsai/pfnstudio),
+the toolkit for training [prior-fitted foundation models](https://arxiv.org/abs/2112.10510).
+
+## Install
+
+```bash
+pip install priorstudio
+```
+
+For training (requires PyTorch):
+
+```bash
+pip install "priorstudio[torch]"
+```
+
+## Commands
+
+```text
+priorstudio init <dir>            # scaffold a new FM project
+priorstudio validate <path>       # check artifacts against JSON Schema
+priorstudio lint <project>        # cross-reference + style checks
+priorstudio sample <prior.yaml>   # draw N tasks from a prior
+priorstudio run <run.yaml>        # execute a training run end-to-end
+priorstudio predict <run-dir>     # inference against a trained checkpoint
+priorstudio export <project>     # tar-gzipped project archive
+```
+
+Run `priorstudio --help` for the full list and `<cmd> --help` for each
+subcommand's flags.
+
+## What this CLI is for
+
+PFN Studio organises every PFN project around five first-class
+artifacts: **priors** (synthetic data generators), **models** (block
+compositions), **evals** (benchmarks + metrics), **runs** (training
+manifests), and **initiatives** (research workstreams). This CLI
+operates on the file layout those artifacts produce — scaffolding new
+projects, validating them, running training, and exporting them for
+sharing.
+
+The full story (concepts, architecture, examples, marketplace catalog)
+lives at the main repo:
+**[github.com/profitopsai/pfnstudio](https://github.com/profitopsai/pfnstudio)**
+
+## License
+
+Apache-2.0.

pfnstudio-0.7.0/pfnstudio/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.4.0"

pfnstudio-0.7.0/pfnstudio/_bundled/.gitignore

@@ -0,0 +1,4 @@
+# Populated at build time by ../../scripts/sync-bundled.sh.
+# The canonical sources live at the repo root (schemas/, templates/).
+*
+!.gitignore

pfnstudio-0.7.0/pfnstudio/_bundled/schemas/eval.schema.json

@@ -0,0 +1,53 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://priorstudio.dev/schemas/eval.schema.json",
+  "title": "Eval",
+  "description": "A benchmark configuration. Reproducible across runs.",
+  "type": "object",
+  "required": ["id", "name", "version", "task", "dataset", "metrics"],
+  "properties": {
+    "id": { "type": "string", "pattern": "^[a-z0-9_-]+$" },
+    "name": { "type": "string" },
+    "version": { "type": "string", "pattern": "^\\d+\\.\\d+\\.\\d+$" },
+    "task": {
+      "type": "string",
+      "enum": ["discovery", "estimation", "rca", "forecast", "classification", "custom"]
+    },
+    "dataset": {
+      "type": "object",
+      "required": ["name"],
+      "properties": {
+        "name": { "type": "string" },
+        "source": { "type": "string", "description": "URL or local path." },
+        "version": { "type": "string" },
+        "split": { "type": "string", "enum": ["train", "val", "test", "full"] }
+      }
+    },
+    "metrics": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": ["name"],
+        "properties": {
+          "name": { "type": "string", "description": "e.g. 'shd', 'pehe', 'auroc'." },
+          "higher_is_better": { "type": "boolean" },
+          "description": { "type": "string" }
+        }
+      }
+    },
+    "baselines": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": ["name"],
+        "properties": {
+          "name": { "type": "string" },
+          "score": { "type": "number" },
+          "source": { "type": "string" }
+        }
+      }
+    },
+    "citations": { "type": "array", "items": { "type": "string" } }
+  }
+}

pfnstudio-0.7.0/pfnstudio/_bundled/schemas/initiative.schema.json

@@ -0,0 +1,37 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://priorstudio.dev/schemas/initiative.schema.json",
+  "title": "Initiative",
+  "description": "A research workstream. Stored as a markdown file with this frontmatter.",
+  "type": "object",
+  "required": ["id", "title", "status", "version_target"],
+  "properties": {
+    "id": {
+      "type": "string",
+      "pattern": "^[0-9]{4}-[a-z0-9-]+$",
+      "description": "Numeric prefix + slug, e.g. '0001-long-range-lags'."
+    },
+    "title": { "type": "string" },
+    "status": {
+      "type": "string",
+      "enum": ["proposed", "in_progress", "blocked", "done", "abandoned"]
+    },
+    "version_target": {
+      "type": "string",
+      "pattern": "^v\\d+\\.\\d+$",
+      "description": "Which roadmap version this lands in."
+    },
+    "owner": { "type": "string" },
+    "depends_on": {
+      "type": "array",
+      "items": { "type": "string" },
+      "description": "IDs of other initiatives."
+    },
+    "tags": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+    "created": { "type": "string", "format": "date" },
+    "updated": { "type": "string", "format": "date" }
+  }
+}

pfnstudio-0.7.0/pfnstudio/_bundled/schemas/model.schema.json

@@ -0,0 +1,43 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://priorstudio.dev/schemas/model.schema.json",
+  "title": "Model",
+  "description": "An architecture config. Composable from blocks in core/.",
+  "type": "object",
+  "required": ["id", "name", "version", "blocks"],
+  "properties": {
+    "id": { "type": "string", "pattern": "^[a-z0-9_-]+$" },
+    "name": { "type": "string" },
+    "version": { "type": "string", "pattern": "^\\d+\\.\\d+\\.\\d+$" },
+    "description": { "type": "string" },
+    "blocks": {
+      "type": "array",
+      "description": "Ordered architecture blocks. Each names a registered block type and its config.",
+      "items": {
+        "type": "object",
+        "required": ["type"],
+        "properties": {
+          "type": { "type": "string", "description": "Block type, e.g. 'temporal_encoder', 'causal_attention', 'treatment_head'." },
+          "name": { "type": "string", "description": "Optional alias for referencing this block elsewhere." },
+          "config": { "type": "object", "additionalProperties": true }
+        }
+      }
+    },
+    "input_shape": { "type": "string", "description": "Symbolic, e.g. '(B, T, D)'." },
+    "output_heads": {
+      "type": "array",
+      "items": {
+        "type": "object",
+        "required": ["name", "task"],
+        "properties": {
+          "name": { "type": "string" },
+          "task": {
+            "type": "string",
+            "enum": ["discovery", "estimation", "rca", "forecast", "classification", "custom"]
+          }
+        }
+      }
+    },
+    "citations": { "type": "array", "items": { "type": "string" } }
+  }
+}

pfnstudio-0.7.0/pfnstudio/_bundled/schemas/prior.schema.json

@@ -0,0 +1,86 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://priorstudio.dev/schemas/prior.schema.json",
+  "title": "Prior",
+  "description": "A synthetic data generator for a prior-fitted foundation model. Each prior has this YAML spec plus a Python implementation in the same directory.",
+  "type": "object",
+  "required": ["id", "name", "version", "kind", "parameters", "outputs"],
+  "properties": {
+    "id": {
+      "type": "string",
+      "pattern": "^[a-z0-9_-]+$",
+      "description": "Stable identifier, kebab/snake-case. Must match directory name."
+    },
+    "name": { "type": "string" },
+    "version": {
+      "type": "string",
+      "pattern": "^\\d+\\.\\d+\\.\\d+$",
+      "description": "Semver. Bump major when sampling distribution changes."
+    },
+    "kind": {
+      "type": "string",
+      "enum": ["scm", "tabular", "temporal", "temporal_causal", "graph", "custom"],
+      "description": "What family of structure the prior generates."
+    },
+    "description": { "type": "string" },
+    "parameters": {
+      "type": "object",
+      "description": "Hyperparameters of the prior. Keys are parameter names; values describe sampling.",
+      "additionalProperties": {
+        "type": "object",
+        "required": ["type"],
+        "properties": {
+          "type": {
+            "type": "string",
+            "enum": ["int", "float", "categorical", "bool"]
+          },
+          "range": {
+            "type": "array",
+            "minItems": 2,
+            "maxItems": 2,
+            "description": "[min, max] for int/float."
+          },
+          "choices": {
+            "type": "array",
+            "description": "Values for categorical."
+          },
+          "default": {},
+          "description": { "type": "string" }
+        }
+      }
+    },
+    "outputs": {
+      "type": "object",
+      "description": "Shape of one sample from the prior.",
+      "required": ["variables"],
+      "properties": {
+        "variables": {
+          "type": "array",
+          "items": {
+            "type": "object",
+            "required": ["name", "type"],
+            "properties": {
+              "name": { "type": "string" },
+              "type": {
+                "type": "string",
+                "enum": ["scalar", "tensor", "graph", "matrix"]
+              },
+              "shape": { "type": "string", "description": "Symbolic shape, e.g. '(T, D)'." },
+              "description": { "type": "string" }
+            }
+          }
+        }
+      }
+    },
+    "citations": {
+      "type": "array",
+      "items": { "type": "string" },
+      "description": "BibTeX keys from literature/references.bib."
+    },
+    "implementation": {
+      "type": "string",
+      "description": "Path to Python module (relative to prior directory).",
+      "default": "prior.py"
+    }
+  }
+}

pfnstudio-0.7.0/pfnstudio/_bundled/schemas/run.schema.json

@@ -0,0 +1,67 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "https://priorstudio.dev/schemas/run.schema.json",
+  "title": "Run",
+  "description": "An experiment manifest. Pins a prior, model, eval, hyperparams, and compute target.",
+  "type": "object",
+  "required": ["id", "prior", "model", "evals", "hyperparams"],
+  "properties": {
+    "id": { "type": "string", "pattern": "^[a-z0-9_-]+$" },
+    "description": { "type": "string" },
+    "prior": {
+      "type": "object",
+      "required": ["id", "version"],
+      "properties": {
+        "id": { "type": "string" },
+        "version": { "type": "string" },
+        "overrides": { "type": "object", "description": "Parameter overrides for this run." }
+      }
+    },
+    "model": {
+      "type": "object",
+      "required": ["id", "version"],
+      "properties": {
+        "id": { "type": "string" },
+        "version": { "type": "string" }
+      }
+    },
+    "evals": {
+      "type": "array",
+      "minItems": 1,
+      "items": {
+        "type": "object",
+        "required": ["id", "version"],
+        "properties": {
+          "id": { "type": "string" },
+          "version": { "type": "string" }
+        }
+      }
+    },
+    "hyperparams": {
+      "type": "object",
+      "additionalProperties": true,
+      "description": "Free-form, but conventionally: lr, batch_size, steps, seed, optimizer."
+    },
+    "compute": {
+      "type": "object",
+      "properties": {
+        "target": { "type": "string", "enum": ["local", "vast", "modal", "runpod", "hf_spaces", "custom"] },
+        "gpu": { "type": "string", "description": "e.g. 'A100-80GB', 'H100'." },
+        "num_gpus": { "type": "integer", "minimum": 1 }
+      }
+    },
+    "tracking": {
+      "type": "object",
+      "properties": {
+        "wandb_project": { "type": "string" },
+        "wandb_run_id": { "type": "string" },
+        "hf_repo": { "type": "string" }
+      }
+    },
+    "results": {
+      "type": "object",
+      "additionalProperties": true,
+      "description": "Filled in after the run completes."
+    }
+  }
+}

pfnstudio-0.7.0/pfnstudio/_bundled/templates/fm-project/README.md

@@ -0,0 +1,39 @@
+# {{project_name}}
+
+> {{one_line_description}}
+
+A prior-fitted foundation model project, scaffolded with [PFN Studio](https://github.com/{{org}}/priorstudio).
+
+## Layout
+
+```
+.
+├── ROADMAP.md            versioned plan (v0.1 → v1.0)
+├── initiatives/          one .md per workstream
+├── priors/               YAML spec + Python impl per prior
+├── models/               architecture configs
+├── evals/                benchmark configs
+├── literature/           BibTeX + per-paper summaries
+└── runs/                 experiment manifests
+```
+
+## Workflow
+
+1. **Add a prior.** Each prior is a directory under `priors/` with a `prior.yaml` (spec) and `prior.py` (implementation). See `priors/example_linear_scm/`.
+2. **Add a model.** YAML config under `models/` describing the block composition.
+3. **Add an eval.** YAML config under `evals/` pinning a dataset, metrics, and baselines.
+4. **Define a run.** YAML manifest under `runs/` linking a prior + model + eval + hyperparams + compute target.
+5. **Track it as an initiative.** Add a markdown file under `initiatives/` and link from `ROADMAP.md`.
+
+## Validate
+
+```bash
+priorstudio validate
+priorstudio lint              # cross-reference checks
+```
+
+## Run
+
+```bash
+priorstudio run runs/example_run.yaml
+```

pfnstudio-0.7.0/pfnstudio/_bundled/templates/fm-project/ROADMAP.md

@@ -0,0 +1,25 @@
+# {{project_name}} Roadmap
+
+## v0.1 — Bootstrap
+
+**Goal:** first prior, first model, first eval round-trip end-to-end.
+
+- [ ] [0001-define-base-prior](initiatives/0001-define-base-prior.md)
+- [ ] [0002-baseline-architecture](initiatives/0002-baseline-architecture.md)
+- [ ] [0003-baseline-eval](initiatives/0003-baseline-eval.md)
+
+## v0.2 — Iteration
+
+**Goal:** prior families and architecture variants, comparable on a fixed eval suite.
+
+- [ ] Prior variants
+- [ ] Architecture variants
+- [ ] Benchmark suite expansion
+
+## v1.0 — Release
+
+**Goal:** publishable model checkpoint + paper-ready results.
+
+- [ ] Model card
+- [ ] Reproducibility audit
+- [ ] HuggingFace release

pfnstudio-0.7.0/pfnstudio/_bundled/templates/fm-project/evals/example_sachs.yaml

@@ -0,0 +1,37 @@
+id: example_sachs
+name: Sachs protein-signaling benchmark
+version: 0.1.0
+task: discovery
+description: |
+  Causal discovery on the Sachs et al. (2005) protein-signaling dataset.
+  Standard benchmark — 11 nodes, ~17 ground-truth edges depending on accepted
+  consensus graph.
+
+dataset:
+  name: sachs
+  source: https://www.science.org/doi/10.1126/science.1105809
+  version: "2005"
+  split: full
+
+metrics:
+  - name: shd
+    higher_is_better: false
+    description: Structural Hamming Distance to ground-truth DAG.
+  - name: f1
+    higher_is_better: true
+    description: F1 over edge presence.
+
+baselines:
+  - name: PC
+    score: 17
+    source: tsamardinos2019
+  - name: GES
+    score: 18
+    source: tsamardinos2019
+  - name: NOTEARS
+    score: 14
+    source: zheng2018notears
+
+citations:
+  - sachs2005causal
+  - zheng2018notears

pfnstudio-0.7.0/pfnstudio/_bundled/templates/fm-project/initiatives/0001-define-base-prior.md

@@ -0,0 +1,32 @@
+---
+id: 0001-define-base-prior
+title: Define base prior
+status: in_progress
+version_target: v0.1
+owner: tbd
+tags: [prior]
+created: 2026-05-09
+updated: 2026-05-09
+---
+
+# Define base prior
+
+## Motivation
+
+Why this prior exists. What real-world phenomenon it's a synthetic stand-in for. What identifiability properties matter.
+
+## Acceptance criteria
+
+- [ ] `prior.yaml` and `prior.py` in `priors/<id>/`, validated against schema
+- [ ] Sampling reproducible from fixed seed
+- [ ] Documented in `priors/<id>/prior.md` (the *why*)
+
+## Open questions
+
+- Parameter ranges?
+- Identifiability?
+- Computational cost per sample?
+
+## References
+
+- See `literature/references.bib`